Mycotoxins and other secondary metabolites produced by fungi and plants pose significant risks to food safety. Liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) is a powerful technique for the simultaneous quantification of a wide array of these contaminants in food samples. The technique's ability to handle complex matrices with a simple 'dilute and shoot' approach has made it the gold standard. However, the validation of LC-MS/MS methods for mycotoxins in processed grain products such as pasta, biscuits, crackers, and muesli, remains limited. These processed products are considerably different from raw materials due to processing steps, additives, and ingredients, which can affect recovery and matrix effects. This study aimed to address this gap by validating an LC-MS/MS method for the simultaneous quantification of a broad range of mycotoxins and other secondary metabolites in these complex matrices, considering the variability between different brands and products. Existing guidelines on method validation vary in their strictness regarding matrix effects. Some require testing of multiple matrices to account for variability, while others rely on the use of stable isotope-labeled internal standards or proficiency testing data. The approach chosen here employs spiking multiple brands of each food matrix to evaluate method performance and address potential relative matrix effects and differences in extraction efficiency.

Several studies have validated LC-MS/MS methods for mycotoxin analysis in various food matrices. However, comprehensive analyses targeting a broad spectrum of mycotoxins and other secondary metabolites (including pyrrolizidine and tropane alkaloids) in processed grain products are scarce. Existing methods often focus on a limited number of regulated mycotoxins, neglecting the potential co-occurrence of other contaminants. The impact of matrix effects in processed foods is also a significant consideration, as the complexity of these matrices can vary substantially between different brands and products. This study builds upon previous work demonstrating the feasibility of a dilute-and-shoot approach for mycotoxin analysis in raw agricultural commodities, extending this approach to the analysis of more complex processed foods. The existing literature highlights the need for robust, high-throughput methods that can accurately quantify a wide range of contaminants in diverse food matrices while accounting for matrix effects.

The study employed an extended version of a previously published LC-MS/MS method to analyze nearly 900 metabolites (730 quantifiable due to available standards). This expanded method required adjustments to the sMRM parameters to ensure data quality. The dwell time was increased to improve signal acquisition, allowing for at least 10 data points per peak. The method included pyrrolizidine and tropane alkaloids, contaminants with similar contamination pathways to mycotoxins. The study used a "dilute and shoot" approach, where samples were extracted with acetonitrile/water/acetic acid (79:20:1, v/v/v), diluted, and directly injected into the LC-MS/MS system without further cleanup. Seven different brands of each matrix (pasta, biscuits, crackers, and muesli) were selected to represent the variability within each food type. Apparent recoveries (RA), matrix effects (SSE), and extraction recoveries (RE) were determined using spiked samples. Repeatability (RSDr) was assessed using seven different samples per matrix, while intermediate precision (RSDWLR) was evaluated with seven technical replicates of a single sample per matrix analyzed over seven weeks. Trueness was evaluated through participation in a proficiency testing program. Limits of quantification (LOQ) were determined according to EURACHEM guidelines. A total of 157 real-world samples from the European market were analyzed to assess the method’s applicability to commercially available products. The LC-MS/MS system used was a QTrap 5500 coupled to a 1290 UHPLC system, employing a Gemini C18 column. A binary methanol/water gradient was utilized for chromatographic separation. External calibration was performed using neat solvent standards. Linearity was assessed by constructing 1/x weighted calibration curves.

The results demonstrated that the dilute-and-shoot approach is applicable to the analysis of processed grain matrices with minimal matrix effects. For 85-88% of the analytes, the apparent recoveries (RA) were within the target range of 70-120%, and 92-94% of the extraction recoveries (RE) met this criterion. Only 7-14% of the analytes exhibited significant matrix effects (±20% deviation). The repeatability (RSDr) was excellent, with over 70% of analyte/matrix combinations showing RSDr <10% and 95-98% complying with the RSDr <20% criterion. Intermediate precision (RSDWLR) was slightly better, with 98.9-99.2% of analytes meeting the <20% criterion. The expanded measurement uncertainty (U_r,exp) was compliant with the <50% criterion for 95-98% of the analytes. Proficiency testing showed satisfactory trueness for the method, with most z-scores between -1 and 1. LOQs were determined for biscuits and muesli, with most values below regulatory limits for mycotoxins in cereal-based products. Analysis of 157 real-world samples confirmed the method's performance and revealed the presence of deoxynivalenol and enniatins in most samples. Only a few samples exceeded the regulatory limit for the sum of ergot alkaloids. The study also showed that extraction efficiencies and matrix effects were largely independent of analyte concentration, indicating that extensive LOQ determination for each matrix is not necessary.

The findings confirm the robustness of the LC-MS/MS dilute-and-shoot method for the high-throughput quantification of a wide range of mycotoxins and other secondary metabolites in complex processed grain matrices. The unexpectedly low matrix effects in processed foods compared to raw commodities suggest that processing may alter or reduce matrix constituents responsible for ionization suppression or enhancement. The excellent repeatability and intermediate precision, despite the variability between different brands, underscores the method's reliability. The results of the real-world sample analysis highlight the widespread co-occurrence of mycotoxins and other natural toxins in commercially available grain products. Although most samples complied with EU regulations, exceeding the ergot alkaloid limit in a few samples emphasizes the importance of continued monitoring. The findings support the use of a matrix-independent estimation of LOQs for broad multi-analyte methods. Future research could explore the impact of different processing techniques on matrix effects and further expand the scope of the method to include additional contaminants.

This study successfully validated a high-throughput LC-MS/MS method for the simultaneous quantification of over 700 mycotoxins and other secondary metabolites in processed grain products. The method's robustness and wide applicability are demonstrated by its performance across different brands and matrices and its ability to detect a wide range of contaminants. Future work should explore additional matrices and contaminants, further investigate the influence of food processing on matrix effects, and examine the combined health effects of the low-level co-occurrence of natural toxins.

While the study included seven different brands per matrix to represent variability, it may not fully capture the range of potential matrix effects found in all processed grain products. The study focused on European market samples, and the generalizability to other regions or product types may be limited. Additionally, the lack of true blank samples for some analytes affected the assessment of matrix effects. While proficiency testing data supported the trueness of the method, it does not eliminate the possibility of unforeseen interferences or biases.